Performance of management procedures for British Columbia Pacific Herring (Clupea pallasii) in the presence of model uncertainty : closing the gap between precautionary fisheries theory and practice

Abstract

The method of setting catch limits for Pacific Herring (Clupea pallasii) fisheries in British Columbia (BC) is similar to precautionary harvest policies found elsewhere in the world; however, 3 out of 5 herring fisheries have been closed in most years since 2006 due to persistent low spawning abundances and low productivity. Although the mechanisms underlying declines of these herring stocks remain unknown, temporal variation in natural mortality and stock assessment over-estimation of abundance are potential factors involved in these outcomes. We used closed-loop simulations to evaluate management procedure (MP) performance for West Coast Vancouver Island (WCVI) and Strait of Georgia (SOG) herring fisheries given uncertainties about past and future herring natural mortality and stock assessment estimation errors. This work represents the first phase of management strategy evaluation under Pacific Herring Renewal, where emphasis is on evaluating current MPs and modifications to these MPs, and not on identifying or selecting the most acceptable MP. We develop three operating models representing hypotheses for how stock-specific natural mortality changes over time. The first model (constant-M) assumes that natural mortality has remained constant over the 1951-2017 period, while the alternative model (time-varying-M) allows natural mortality to vary over that time. The time-varying-M operating model is further divided into two models for projecting future patterns in natural mortality. A density-independent-M model assumes that future natural mortality rates will fluctuate randomly around the recent 10-year average, while a density-dependent-M model allows random pulses of high natural mortality when spawning biomass is low. Increasing natural mortality rates are of concern and are relevant given increasing predator biomass. We simulated performance of nine feedback harvest control rules (HCRs) given by combinations of maximum harvest rate (20% vs 10%), HCR form (i.e., hockey-stick vs. minimum escapement), operational control points defining biomass cutoffs (25%, 30%, and 50% of B0) and thresholds below which harvest rates are reduced (none vs. 60% of B0), and absolute catch caps (0 vs 2,000 t for WCVI and 0 vs. 30,000 t for SOG). For WCVI, results show that the current MP would fail to meet spawning biomass objectives under most operating models. Reducing the maximum harvest rate from 20% to 10% and capping fishery quotas at a maximum 2,000 t would reduce the effective harvest rate and protect against over-estimates of abundance when they occur, thus providing acceptable performance against biomass objectives for two of three operating models. For SOG herring, the current MP was robust across almost all scenarios and objectives we examined (thus a more restrictive cap was not explored). For both WCVI and SOG herring, the maximum target harvest rate was the most important harvest control rule element controlling management performance compared to the shape and/or operational control points in harvest control rules.